Telephones and other communications devices for sending audio information through the public switched telephone network (PSTN) have existed for a substantial period of time and are well understood. Recently, however, telephones have been created that exchange audio information through packets switched networks such as the Internet. These telephones use a digitized format that breaks down audio information into discrete Internet protocol (IP) packets. These voice over IP (VoIP) are sent and received through the packet switched network.
Exchanging audio information through a packet switched network presents a variety of problems. For example, errors in the packet transmission may result in echo, scratchy audio, dropped calls, delay and/or jitter in the audio string. As long as the fullest quality of VoIP is perceived to be inferior to the voice quality of traditional telephony systems, users will remain reluctant to employ VoIP systems.
Packet replication is a known technique for reducing the probability of packet loss and for reducing jitter associated with packet time of arrival. Packet replication may be invoked at the edge of a communication network where voice packets enter or exit the network. As the number of replication points increases, the load in the network increases, thus, adversely impacting the overall performance of the network. Points in the network where replication takes place are static.
As a given packet is replicated and sent via different communication paths, the resulting voice quality increases up through a certain number of duplications. Continuing to replicate the voice packets beyond this number will increase congestion in the network while providing only marginal improvements and voice quality, if any.